A Humanoid Robot Ran 100 Metres in 9.39 Seconds
Tiangong Ultra finished a 100-metre robot race in 9.39 seconds, faster than the human world record and less than half its own 2025 time. The jump in speed is real, but the machine stopped by hitting padding and the contest did not test safe movement around people.
A humanoid robot named Tiangong Ultra covered 100 metres in 9.39 seconds at Beijing's World Humanoid Robot Games on August 22. The number is faster than Usain Bolt's 9.58-second human record, but the machines finished by crashing into padded barriers rather than slowing safely.
The 30-second summary
- What happened? Tiangong Ultra won a 100-metre robot race in 9.39 seconds, while Honor's Lightning finished in 9.47 seconds.
- Why does it matter? Tiangong took 21.50 seconds over the same distance at last year's games, so the result shows a sharp gain in humanoid running speed.
- What is the catch? This was a robot competition, not a human athletics record. The machines were built for speed, and both needed barriers to stop.
KEY NUMBER
Tiangong Ultra cut its 100-metre time from 21.50 seconds in 2025 to 9.39 seconds in 2026.
What happened on the track
Reuters reported that Tiangong Ultra overtook Lightning near the finish. Lightning had led early, then completed the race in 9.47 seconds. A preparatory test had put Lightning at 9.32 seconds and a peak speed of 14.5 metres per second.
The event's published competition rules required ordinary track races to be autonomous. The organisers exempted only the 100-metre and 400-metre hurdle events, not the flat sprint.
Why the time is impressive but not a human record
The comparison with Bolt is useful because almost everyone understands 100 metres. It is not a like-for-like sporting record. Robot makers can change leg length and mechanical proportions, and their machines did not show the controlled deceleration required from a human athlete.
That limitation was visible at the finish. Tiangong veered into the sidelines after striking the stopping mat, while Lightning collapsed and was carried away. The race proved rapid forward motion on a prepared track. It did not test crowd safety, uneven ground, energy efficiency or recovery after a fall.
Before we call this superhuman running
- The 9.39-second result came from a competition demonstration, not an independently standardized engineering benchmark.
- The robot's stopping method would be unacceptable around workers, pedestrians or customers.
- Short sprint speed says little about battery life, useful payload, dexterity or reliable operation over a work shift.
- The games reward a narrow task under known conditions, while real environments contain obstacles and people who move unpredictably.
The harder tests are happening beside the spectacle
The five-day event includes 51 contests and more than 2,000 robots. Some are visual crowd-pleasers, but others ask machines to tighten screws, handle packages, charge electric vehicles or respond to emergencies. Those slower tasks may reveal more about whether humanoids can become useful equipment.
NewTqnia recently examined a different part of that transition when 15 robot traffic officers began working on Hangzhou streets. We also covered a model that combined movement and heart data to classify athletic performance, a reminder that a headline score matters only when the test behind it matches the intended use.
Takeaway
A 9.39-second sprint is a striking measure of progress because the same robot needed more than twice as long one year earlier. The next meaningful milestone is not another tenth of a second. It is a humanoid that can move quickly, stop safely and repeat useful work without a padded wall waiting at the finish.
Verified topics and entities
Sources and citations4 sources
External references used to support the reporting in this article.
Published by
NewTqnia Sports Technology Desk
An institutional editorial team within NewTqnia
